Navigation satellite systems (NSS) provide functionalities to rovers such that the rover is able to determine its position. Examples of a rover include navigations systems, mobile phones and the like. Navigation satellite systems include global navigation satellite systems (GNSS) as for example the Global Positioning System (GPS), the GLONASS system, the Galileo system, the BeiDou system and others. Navigation satellite systems may however not necessarily cover the complete globe and can be restricted to a certain area or region on earth.
A technique known as Real-Time Kinematic (RTK) positioning is commonly used to improve the positioning accuracy in a NSS. RTK uses one or more reference stations at fixed locations on the surface of the earth. A reference station continuously measures or otherwise obtains its distance from various satellites. The measured distance is compared with the known distance between the reference station and the satellite which can be modeled based on the position of the reference station and the orbit of the satellite. The difference between the measured distance and the modelled distance are transmitted to nearby rover stations. The rover stations can then use the received difference to improve its positioning. With this technique, a positioning accuracy in the order of centimeters can be reached.
Alternatively, the satellite measurements and reference station position used to compute these corrections can be transmitted to rover stations, allowing various methods at the rover to be used to improve its position. Although the transmitted data in this case do not comprise actual corrections, the term Real-Time Corrections (RTC) is commonly used in the GNSS industry to describe both types of transmitted data. Hereafter, Real-Time Corrections (RTC) shall be used to describe a combination of satellite measurements and reference station positions.
Real-Time Corrections (RTC) that are delivered to one or more rover stations can be generated by a single reference station, a network of local reference stations, or a wide-area or global network of reference stations. When a plurality of reference stations is used, these may be connected to a computer server which delivers the Real-Time Corrections (RTC) to the rover stations.
For the purposes of the description, it can be assumed that the RTC stream applies to a single reference station.
The Real-Time Correction (RTC) data offered by the reference station enables many of the systematic errors affecting the rover to be removed or reduced. For example, the rover measurements can be affected by error sources such as satellite clock, satellite orbit, ionosphere bias, tropospheric bias. The satellite clock errors vary over time. Ionospheric and tropospheric errors have both spatial and temporary variation.
The data transmission from the reference station to the rover station can be performed via data links such as radio/modem, cellular phone networks, satellite radio communications, or the internet. This data transmission is preferably performed in real time. When a plurality of reference stations is connected to a computer server, the server may complete the connection to the rover stations using any of the said data links. When the internet is used, the data link may use internet protocols such as TCP-IP, UDP or PPP.
The bandwidth of these data links is of crucial importance when performing high-precision, real-time NSS positioning. For example, many radio/modems have an effective bandwidth throughput of 9,600 bits per second. Satellite based data links are very expensive to lease and therefore it is advantageous to limit the correction information sent via satellites. Also data transmission via a cellular phone network is expensive. Therefore, the data size of the correction data should be kept to a minimum.
Furthermore, many data links are not 100% reliable, i.e. they are subject to outages and bit errors. Therefore, the method used for transmitting the correction data from a reference station to the rover station should have a certain robustness with respect to outages of the data link.
In 2013, there are currently approximately 75 navigation satellites in space including 31 GPS satellites, 24 GLONASS (Russia) satellites, 14 BeiDou (BDS) (Chinese) satellites, 4 Galileo (EU) satellites and 1 QZSS (Japanese) satellite. The number of satellites will grow over the next decades. Furthermore, new navigation satellites will broadcast on 3 frequency bands, rather than just 2. The expansion of satellite count and satellite signals will lead to a natural increase in the amount of information that needs to be distributed in the RTC stream for high-precision, real-time NSS positioning applications.
In view of the above, a reference correction data format should consider:                size: The amount of data sent should be minimized without compromising the precision of the content. Small data content leads to a reduction in correction latency;        robustness: The format should be resilient to bit errors on the data stream. A single lost packet should not cause a flow on loss of further information;        extendibility: Corrections for new satellites systems and signals should be readily included in the data format; and        simplicity: the encode/decode algorithms should be as simple and computationally efficient as possible.        
There are some prior art correction formats for high precision NSS applications. One example is the compact measurement record (CMR) format which is for example described in Nick Talbot, “Compact Data Transmission Standard for High-Precision GPS”, proceedings of ION-GPS 96, Kansas City, 1996.
Spatial Data Compression
The CMR format is based around a spatial data compression algorithm. NSS pseudorange measurements have a range of 20,000 to 25,000 kilometers with a required precision of around 1 centimeter. Instead of sending out the full pseudorange measurement, the pseudorange modulo 1 light millisecond (about 300 kilometer) is sent. This spatial data compression algorithm greatly reduces the number of bits required to transmit the pseudorange data. The ambiguity in the pseudorange data is readily reconciled by the approximate ordinates of the reference station and the coordinates of each satellite obtained from the broadcast orbits. This is an example of compressing the satellite measurements rather than generating corrections, although the term Real-Time Corrections (RTC) is commonly used in the GNSS industry for all types of compression methods.
NSS carrier phase measurements have millimeter level precision and similar range to NSS pseudorange measurements. In the CMR format, NSS carrier phase measurements are transmitted as offsets relative to the respective satellite pseudorange measurement. This approach again reduces the range of the transmitted parameters. Likewise, carrier phase and pseudorange measurements on the secondary frequency band (for example in GPS the L2 frequency) are sent as offsets relative to the primary (e.g. GPS L1) frequency band.
The CMR format has been further improved to the CMRx format (see for example U.S. Pat. No. 8,044,849). The CMRx format includes more complex spatial data compression algorithms that deliver a very high level of spatial data compression. The CMRx format exploits the intrinsic properties of the NSS observations. For example, the tropospheric delay of the NSS signals is accurately accounted for using a standard (simplified Hopfield) model.
The user-satellite range is by far the largest component of NSS measurement that needs to be encoded/decoded in real-time, high-precision NSS positioning applications. NSS satellite locations are known to approximately 10 meters from the respective broadcast ephemerides. The location of the transmitting reference station is known and broadcast as part of the CMRx message. Therefore, the combination of reference station location and the NSS orbital information allows the user satellite-range to be computed at reference stations and rover stations. Instead of using the full user satellite range, just a portion of the measurements needs to be sent.
FIG. 1 shows this compression by illustrating a single satellite being tracked by a reference receiver. The position of the satellite is known to an accuracy of around 10 meters (orbit uncertainty) using the orbit information broadcast by the satellite. Furthermore, the reference receiver location is also known to a very high precision. Therefore, the distance between the reference receiver and the satellite is known to high precision. Thus, the carrier phase measurements do not need to be transmitted in full but can be transmitted modulo an ambiguity window representing a range of a particular size (e.g. 100 meters). The size of the ambiguity window needs to be carefully considered. A too small window size may result in an ambiguity in the broadcast carrier phase measurement. A too large window size increases the magnitude of the quantities that need to be encoded.
A major advantage of the spatial data compression scheme is that it is insensitive to changes in the broadcast NSS ephemerides parameters. For example, GPS satellites broadcast a new orbit and clock information every few hours. Each update of the broadcast ephemerides is marked by a change to the issue of data ephemeris (IODE) flag in the data. Normally each update causes up to meter level jumps in the computed (modelled) satellite locations. Some data formats handle IODE changes by including the IODE flag in the data.
Temporal Data Compression
NSS satellites today move in an inclined geo synchronous orbit (IGEO), an equatorial geo synchronous (GEO) or medium earth orbit (MEO). The apparent angular rate of the satellites is relatively low (around 0.5 degrees per minute for MEO). Atmospheric errors are also relatively slowly changing. For example under stable ionospheric conditions, the ionospheric bias typically changes by 1-10 millimeters/second (mm/s) for rising MEO satellites and 1-5 mm/s for satellites overhead. Similarly, tropospheric error rates are on the order of 10 mm/s for rising satellites and a few mm/s for satellites overhead.
Furthermore, the acceleration of the satellites with respect to the user-satellite range is quite constant and readily modeled over time. The relatively constant satellite acceleration means that errors in the reference station coordinates propagate as slow variation in the reference station-to-satellite computed range (˜a few mm/s for an error in the reference station coordinates of say 2 m). Similarly, satellite ephemeris errors, for a given issue of data, cause slow variation in the computed ranges (<1 mm/s).
Because of the intrinsic stability of the satellite atomic frequency standards, satellite clock error growth is generally around 1-2 mm/s.
Satellite signals can be reflected off objects surrounding the receiving antenna, thus inducing signal multipath errors. Multipath errors for a station receiver often exhibit correlation times between 5 and 120 s.
In view of these considerations, temporal compression exploits the slow time variation of NSS measurements. Normally temporal compression algorithms divide transmissions into major and delta epochs. The major epochs (reference epochs) contain fully reconstructable NSS observations. The delta epochs are referred back to the last major epoch. FIG. 2 shows an example of a temporal compression. In the example shown in FIG. 2, four delta messages are transmitted between two reference messages.
The advantage of the major/delta temporal compression scheme is that the average size of the transmitted data is reduced. However, FIG. 3 shows a disadvantage of the temporal compression scheme. FIG. 3 shows that the data packet size peaks at the time of transmission of the major messages. Thus, the data transmission is not evenly distributed over time. This can cause problems if the peaks exceed the data link bandwidth even though the average data packet size is below the data link bandwidth. Furthermore, the temporal data compression scheme has a weakness in that the loss of a single major packet leads to the subsequent loss of dependent delta messages. This is particularly problematic when the data link is unreliable. Furthermore, some techniques require issue of data ephemeris information in messages which adds complexity.
In view of the above, the present invention aims at providing a technique to reduce the amount of correction data transmitted in RTK applications without degrading the robustness with respect to data link outages.